Access Network Measurements Source: - - PowerPoint PPT Presentation

J.Fabini RAIM 2015: Measurement of Broadband Access Networks 1

Access Network Measurements

Delay = d0 + PacketSize / Capacity?

• Access networks are session stateful
• Packet-based processing
• Pronounced asymmetry

uplink vs. downlink

• Cellular Broadband Networks
• On-demand capacity allocation

[3] Fabini et al.: „RDM: Facing the Challenge of Modern Networks“, doi:10.4108/icst.Valuetools.2014.258181

Source: http://commons.wikimedia.org/wiki/ File%3ALautsprecherkabel_Makro_nah.jpg

J.Fabini RAIM 2015: Measurement of Broadband Access Networks 2

On-Demand Capacity Allocation

(a) HSPA low avg. bit rate (b) HSPA higher avg bit rate

• Observations:
• Stream patterns matter! (application, measurements, …)
• Packet inter-departure time, average bit rate, packet size, …
• Packet start-time matters (details at HOPS-NG IRTF meeting)!
• Start-time randomness imperative for unbiased results!

J.Fabini RAIM 2015: Measurement of Broadband Access Networks 3

Delay vs. Capacity Reasoning

(a) VDSL UL (768 kbit/s) (b) VDSL DL (8 Mbit/s)

• Observations:
• Delay not necessarily (inverse) proportional to link capacity!
• Reason in this case: interleaving activated on VDSL downlink

[1] Fabini and Zseby: „M2M communication delay challenges: Application and measurement perspectives“, doi: 10.1109/I2MTC.2015.7151564

J.Fabini RAIM 2015: Measurement of Broadband Access Networks 4

Conclusion

• Networks and middleboxes bias on communications
• At low load, when operating within specifications
• State vs. Coding
• HSDPA 384k UL

middlebox?

• Access networks: middleboxes
• “Dormant” middlebox:

time-domain vs. value-domain

• Main conclusion:

Think (at least) twice before abstracting access link behavior

Contact: Joachim.Fabini@tuwien.ac.at

Source: http://commons.wikimedia.org/wiki/ File%3ALautsprecherkabel_Makro_nah.jpg

J.Fabini RAIM 2015: Measurement of Broadband Access Networks 5

[1] Fabini and Morton: IETF RFC 7312 “Advanced Stream and Sampling Framework for the IPPM” [2] Fabini and Abmayer: “Delay Measurement Methodology Revisited: Time-slotted Randomness Cancellation“, doi:10.1109/TIM.2013.2263914 [3] Fabini et al.: „RDM: Facing the Challenge of Modern Networks“, doi:10.4108/icst.Valuetools.2014.258181 [4] Fabini and Zseby: „M2M communication delay challenges: Application and measurement perspectives“, doi: 10.1109/I2MTC.2015.7151564 [5] Fabini and Zseby: „The Right Time: Reducing Effective End-to-End Delay in Time- Slotted Packet-Switched Networks“, doi:10.1109/TNET.2015.2451708

Bibliography

Contact: Joachim.Fabini@tuwien.ac.at